Registration & Submission Deadline
Wednesday, April 10, 2013 @ 11:59 PM
April 23rd & 24th, 2013Claire T. Carney Library Addition
RBF-based Image Compression
Jacob J. Sousa
Image compression is typically done using some method based on the discrete cosine transform (DCT, related to the Fourier transform), but there is also the possibility of image compression based on radial basis function (RBF) interpolation. Rather than representing an image as a combination of cosine functions of varying frequencies as DCT methods do, an RBF method uses radial functions (functions whose value is dependent only on the distance from some point) centered at varying points on the image. While research has been done in this area before, ultimately there haven't been very many attempts to create an RBF-based image compression scheme and so it continues to be worth investigating. This project primarily concerns methods for choosing an appropriate set of basis functions to allow for reasonably accurate reconstruction of the image; other elements (how to efficiently encode the resulting set of functions/weights, efficiently reconstructing the image afterward), while important, are not the focus.